Intellectual Property India Publishes Patent Application for 'Fraud Detection Systems For Banking Transactions Using Machine Learning Algorithms' Filed by Parthasarathi P; Pavithra Sri S; Sara Maria Priscilla A; Sunmitha S J; and Rajeshkumar G

MUMBAI, India, April 10 -- Intellectual Property India has published a patent application (202441076399 A) filed by Parthasarathi P; Pavithra Sri S; Sara Maria Priscilla A; Sunmitha S J; and Rajeshkumar G, Namakkal, Tamil Nadu, on Oct. 9, 2024, for 'fraud detection systems for banking transactions using machine learning algorithms.'

Inventor(s) include Parthasarathi P; Pavithra Sri S; Sara Maria Priscilla A; Sunmitha S J; and Rajeshkumar G.

The application for the patent was published on April 10, under issue no. 15/2026.

According to the abstract released by the Intellectual Property India: "With an emphasis on identifying fraudulent activity in real-time across financial systems, including UPI and general type transactions, we offer a reliable and fault-tolerant pipeline architecture designed for fraud detection in banking transactions. The overall efficacy of fraud detection is increased by our suggested method, which combines several machine learning algorithms into a hierarchical structure to provide redundancy, accuracy, and reliability. The pipeline starts with transactional data that is gathered from financial systems and goes through several phases of processing, such as risk assessment, fraud mitigation techniques, and anomaly detection. The design includes fallback methods, which allow another pre-configured algorithm to effortlessly take over in the case that one algorithm fails, in order to provide fault tolerance.The system uses algorithms like Random Forest and XGBoost for anomaly identification, with the ability to switch between them depending on performance circumstances. Real-time replies are ensured by the system's dynamic application of several detection techniques to identify fraudulent patterns. Furthermore, fraud verification uses isolation forests and clustering algorithms to provide reliable and ongoing fraud detection even in complicated situations. The pipeline's last phase includes adaptive learning and fraud protection, which enable the system to gradually get better at making decisions. The suggested multi-layered design strengthens the resilience of fraud detection systems by offering a dependable and expandable framework that can preserve operational integrity in the banking industry under a range of circumstances."

Disclaimer: Curated by HT Syndication.